@kadoa/node-sdk 0.18.1 → 0.19.1

package/dist/index.d.mts

@@ -377,19 +377,10 @@ interface MonitoringConfig {
      */
     'channels'?: Array<any | null>;
     /**
-     * Array of conditions
+     * Condition group with logical operator and conditions
      */
-    'conditions'?: Array<null>;
-    /**
-     * Logical operator for conditions
-     */
-    'logicalOperator'?: MonitoringConfigLogicalOperatorEnum;
+    'conditions'?: any;
 }
-declare const MonitoringConfigLogicalOperatorEnum: {
-    readonly And: "AND";
-    readonly Or: "OR";
-};
-type MonitoringConfigLogicalOperatorEnum = typeof MonitoringConfigLogicalOperatorEnum[keyof typeof MonitoringConfigLogicalOperatorEnum];
 
 /**
  * Kadoa API
@@ -648,6 +639,22 @@ interface AgenticWorkflow {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * Natural language instructions for the AI agent (10-5000 characters). Describe what data to extract and how to navigate the site
      */
@@ -1168,6 +1175,22 @@ interface WorkflowWithEntityAndFields {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * Entity name for extraction (e.g., \'Product\', \'JobListing\')
      */
@@ -1182,6 +1205,7 @@ declare const WorkflowWithEntityAndFieldsNavigationModeEnum: {
     readonly PaginatedPage: "paginated-page";
     readonly PageAndDetail: "page-and-detail";
     readonly AgenticNavigation: "agentic-navigation";
+    readonly AllPages: "all-pages";
 };
 type WorkflowWithEntityAndFieldsNavigationModeEnum = typeof WorkflowWithEntityAndFieldsNavigationModeEnum[keyof typeof WorkflowWithEntityAndFieldsNavigationModeEnum];
 declare const WorkflowWithEntityAndFieldsIntervalEnum: {
@@ -1276,6 +1300,22 @@ interface WorkflowWithExistingSchema {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * ID of an existing schema configuration. Use this to reference a previously defined schema without re-defining extraction fields
      */
@@ -1286,6 +1326,7 @@ declare const WorkflowWithExistingSchemaNavigationModeEnum: {
     readonly PaginatedPage: "paginated-page";
     readonly PageAndDetail: "page-and-detail";
     readonly AgenticNavigation: "agentic-navigation";
+    readonly AllPages: "all-pages";
 };
 type WorkflowWithExistingSchemaNavigationModeEnum = typeof WorkflowWithExistingSchemaNavigationModeEnum[keyof typeof WorkflowWithExistingSchemaNavigationModeEnum];
 declare const WorkflowWithExistingSchemaIntervalEnum: {
@@ -3641,6 +3682,22 @@ interface V4WorkflowsWorkflowIdMetadataPutRequest {
      * Additional static data for the workflow
      */
     'additionalData'?: object;
+    /**
+     * Maximum pages to crawl (only for crawler workflows)
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (only for crawler workflows)
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths (only for crawler workflows)
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths (only for crawler workflows)
+     */
+    'pathsFilterOut'?: Array<string>;
 }
 declare const V4WorkflowsWorkflowIdMetadataPutRequestUpdateIntervalEnum: {
     readonly OnlyOnce: "ONLY_ONCE";
@@ -6020,6 +6077,7 @@ interface CreateWorkflowInput {
     autoStart?: boolean;
     schedules?: string[];
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 declare const TERMINAL_JOB_STATES: Set<JobStateEnum>;
 declare const TERMINAL_RUN_STATES: Set<string>;
@@ -6075,6 +6133,7 @@ interface ExtractionOptionsInternal {
     notifications?: NotificationOptions;
     autoStart?: boolean;
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 type ExtractionOptions = {
     urls: string[];
@@ -6250,7 +6309,7 @@ declare class SchemaBuilder {
 
 interface ExtractOptionsInternal {
     urls: string[];
-    name: string;
+    name?: string;
     description?: string;
     navigationMode: NavigationMode;
     entity: EntityConfig;
@@ -6259,6 +6318,7 @@ interface ExtractOptionsInternal {
     schedules?: string[];
     location?: LocationConfig;
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 interface ExtractOptions extends Omit<ExtractOptionsInternal, "navigationMode" | "entity"> {
     navigationMode?: NavigationMode;
@@ -6277,6 +6337,7 @@ interface PreparedExtraction {
     }) => PreparedExtraction;
     bypassPreview: () => PreparedExtraction;
     setLocation: (options: LocationConfig) => PreparedExtraction;
+    withPrompt: (prompt: string) => PreparedExtraction;
     create: () => Promise<CreatedExtraction>;
 }
 interface RunWorkflowOptions {
@@ -6314,6 +6375,7 @@ declare class ExtractionBuilderService {
     private _jobId;
     private _notificationOptions;
     private _monitoringOptions;
+    private _userPrompt;
     get options(): ExtractOptionsInternal;
     get notificationOptions(): NotificationOptions | undefined;
     get monitoringOptions(): WorkflowMonitoringConfig | undefined;
@@ -6330,6 +6392,7 @@ declare class ExtractionBuilderService {
         schedules: string[];
     }): PreparedExtraction;
     setLocation(options: LocationConfig): PreparedExtraction;
+    withPrompt(prompt: string): PreparedExtraction;
     create(): Promise<CreatedExtraction>;
     waitForReady(options?: WaitForReadyOptions): Promise<GetWorkflowResponse>;
     run(options?: RunWorkflowOptions): Promise<FinishedExtraction>;

package/dist/index.d.ts

@@ -377,19 +377,10 @@ interface MonitoringConfig {
      */
     'channels'?: Array<any | null>;
     /**
-     * Array of conditions
+     * Condition group with logical operator and conditions
      */
-    'conditions'?: Array<null>;
-    /**
-     * Logical operator for conditions
-     */
-    'logicalOperator'?: MonitoringConfigLogicalOperatorEnum;
+    'conditions'?: any;
 }
-declare const MonitoringConfigLogicalOperatorEnum: {
-    readonly And: "AND";
-    readonly Or: "OR";
-};
-type MonitoringConfigLogicalOperatorEnum = typeof MonitoringConfigLogicalOperatorEnum[keyof typeof MonitoringConfigLogicalOperatorEnum];
 
 /**
  * Kadoa API
@@ -648,6 +639,22 @@ interface AgenticWorkflow {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * Natural language instructions for the AI agent (10-5000 characters). Describe what data to extract and how to navigate the site
      */
@@ -1168,6 +1175,22 @@ interface WorkflowWithEntityAndFields {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * Entity name for extraction (e.g., \'Product\', \'JobListing\')
      */
@@ -1182,6 +1205,7 @@ declare const WorkflowWithEntityAndFieldsNavigationModeEnum: {
     readonly PaginatedPage: "paginated-page";
     readonly PageAndDetail: "page-and-detail";
     readonly AgenticNavigation: "agentic-navigation";
+    readonly AllPages: "all-pages";
 };
 type WorkflowWithEntityAndFieldsNavigationModeEnum = typeof WorkflowWithEntityAndFieldsNavigationModeEnum[keyof typeof WorkflowWithEntityAndFieldsNavigationModeEnum];
 declare const WorkflowWithEntityAndFieldsIntervalEnum: {
@@ -1276,6 +1300,22 @@ interface WorkflowWithExistingSchema {
      * Additional data for the workflow (e.g. IDs from your system to link data)
      */
     'additionalData'?: any | null;
+    /**
+     * Maximum pages to crawl (default: 10,000, max: 100,000). Only used when navigationMode is \'all-pages\'
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (default: 50, max: 200). Only used when navigationMode is \'all-pages\'
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths during crawling. Only used when navigationMode is \'all-pages\'
+     */
+    'pathsFilterOut'?: Array<string>;
     /**
      * ID of an existing schema configuration. Use this to reference a previously defined schema without re-defining extraction fields
      */
@@ -1286,6 +1326,7 @@ declare const WorkflowWithExistingSchemaNavigationModeEnum: {
     readonly PaginatedPage: "paginated-page";
     readonly PageAndDetail: "page-and-detail";
     readonly AgenticNavigation: "agentic-navigation";
+    readonly AllPages: "all-pages";
 };
 type WorkflowWithExistingSchemaNavigationModeEnum = typeof WorkflowWithExistingSchemaNavigationModeEnum[keyof typeof WorkflowWithExistingSchemaNavigationModeEnum];
 declare const WorkflowWithExistingSchemaIntervalEnum: {
@@ -3641,6 +3682,22 @@ interface V4WorkflowsWorkflowIdMetadataPutRequest {
      * Additional static data for the workflow
      */
     'additionalData'?: object;
+    /**
+     * Maximum pages to crawl (only for crawler workflows)
+     */
+    'maxPages'?: number;
+    /**
+     * Maximum crawl depth (only for crawler workflows)
+     */
+    'maxDepth'?: number;
+    /**
+     * Regex patterns to include specific paths (only for crawler workflows)
+     */
+    'pathsFilterIn'?: Array<string>;
+    /**
+     * Regex patterns to exclude specific paths (only for crawler workflows)
+     */
+    'pathsFilterOut'?: Array<string>;
 }
 declare const V4WorkflowsWorkflowIdMetadataPutRequestUpdateIntervalEnum: {
     readonly OnlyOnce: "ONLY_ONCE";
@@ -6020,6 +6077,7 @@ interface CreateWorkflowInput {
     autoStart?: boolean;
     schedules?: string[];
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 declare const TERMINAL_JOB_STATES: Set<JobStateEnum>;
 declare const TERMINAL_RUN_STATES: Set<string>;
@@ -6075,6 +6133,7 @@ interface ExtractionOptionsInternal {
     notifications?: NotificationOptions;
     autoStart?: boolean;
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 type ExtractionOptions = {
     urls: string[];
@@ -6250,7 +6309,7 @@ declare class SchemaBuilder {
 
 interface ExtractOptionsInternal {
     urls: string[];
-    name: string;
+    name?: string;
     description?: string;
     navigationMode: NavigationMode;
     entity: EntityConfig;
@@ -6259,6 +6318,7 @@ interface ExtractOptionsInternal {
     schedules?: string[];
     location?: LocationConfig;
     additionalData?: Record<string, unknown>;
+    userPrompt?: string;
 }
 interface ExtractOptions extends Omit<ExtractOptionsInternal, "navigationMode" | "entity"> {
     navigationMode?: NavigationMode;
@@ -6277,6 +6337,7 @@ interface PreparedExtraction {
     }) => PreparedExtraction;
     bypassPreview: () => PreparedExtraction;
     setLocation: (options: LocationConfig) => PreparedExtraction;
+    withPrompt: (prompt: string) => PreparedExtraction;
     create: () => Promise<CreatedExtraction>;
 }
 interface RunWorkflowOptions {
@@ -6314,6 +6375,7 @@ declare class ExtractionBuilderService {
     private _jobId;
     private _notificationOptions;
     private _monitoringOptions;
+    private _userPrompt;
     get options(): ExtractOptionsInternal;
     get notificationOptions(): NotificationOptions | undefined;
     get monitoringOptions(): WorkflowMonitoringConfig | undefined;
@@ -6330,6 +6392,7 @@ declare class ExtractionBuilderService {
         schedules: string[];
     }): PreparedExtraction;
     setLocation(options: LocationConfig): PreparedExtraction;
+    withPrompt(prompt: string): PreparedExtraction;
     create(): Promise<CreatedExtraction>;
     waitForReady(options?: WaitForReadyOptions): Promise<GetWorkflowResponse>;
     run(options?: RunWorkflowOptions): Promise<FinishedExtraction>;

package/dist/index.js

@@ -4427,19 +4427,41 @@ var ExtractionService = class {
       );
     }
     let workflowId;
-    const resolvedEntity = await this.entityResolverService.resolveEntity(
-      options.entity || "ai-detection",
-      {
-        link: config.urls[0],
-        location: config.location,
-        navigationMode: config.navigationMode
+    const isAgenticNavigation = config.navigationMode === "agentic-navigation";
+    if (isAgenticNavigation) {
+      if (!config.userPrompt) {
+        throw new KadoaSdkException(
+          "userPrompt is required when navigationMode is 'agentic-navigation'",
+          {
+            code: "VALIDATION_ERROR",
+            details: { navigationMode: config.navigationMode }
+          }
+        );
       }
-    );
+    }
+    let resolvedEntity;
+    if (isAgenticNavigation) {
+      const entityConfig = options.entity || "ai-detection";
+      resolvedEntity = {
+        entity: typeof entityConfig === "object" && "name" in entityConfig ? entityConfig.name : void 0,
+        fields: typeof entityConfig === "object" && "fields" in entityConfig ? entityConfig.fields : []
+      };
+    } else {
+      resolvedEntity = await this.entityResolverService.resolveEntity(
+        options.entity || "ai-detection",
+        {
+          link: config.urls[0],
+          location: config.location,
+          navigationMode: config.navigationMode
+        }
+      );
+    }
     const hasNotifications = !!config.notifications;
     const workflowRequest = {
       ...config,
       fields: resolvedEntity.fields,
-      ...resolvedEntity.entity !== void 0 ? { entity: resolvedEntity.entity } : {}
+      ...resolvedEntity.entity !== void 0 ? { entity: resolvedEntity.entity } : {},
+      ...config.userPrompt ? { userPrompt: config.userPrompt } : {}
     };
     const result = await this.workflowsCoreService.create(workflowRequest);
     workflowId = result.id;
@@ -4627,20 +4649,46 @@ var ExtractionBuilderService = class {
     this._options.location = options;
     return this;
   }
+  withPrompt(prompt) {
+    assert__default.default(this._options, "Options are not set");
+    this._userPrompt = prompt;
+    this._options.userPrompt = prompt;
+    return this;
+  }
   async create() {
     assert__default.default(this._options, "Options are not set");
     const { urls, name, description, navigationMode, entity } = this.options;
+    const isAgenticNavigation = navigationMode === "agentic-navigation";
+    if (isAgenticNavigation) {
+      if (!this._userPrompt) {
+        throw new KadoaSdkException(
+          "userPrompt is required when navigationMode is 'agentic-navigation'",
+          {
+            code: "VALIDATION_ERROR",
+            details: { navigationMode }
+          }
+        );
+      }
+    }
     const isRealTime = this._options.interval === "REAL_TIME";
     const useSelectorMode = isRealTime && entity === "ai-detection";
-    const resolvedEntity = await this.entityResolverService.resolveEntity(
-      entity,
-      {
-        link: urls[0],
-        location: this._options.location,
-        navigationMode,
-        selectorMode: useSelectorMode
-      }
-    );
+    let resolvedEntity;
+    if (isAgenticNavigation) {
+      resolvedEntity = {
+        entity: typeof entity === "object" && "name" in entity ? entity.name : void 0,
+        fields: typeof entity === "object" && "fields" in entity ? entity.fields : []
+      };
+    } else {
+      resolvedEntity = await this.entityResolverService.resolveEntity(
+        entity,
+        {
+          link: urls[0],
+          location: this._options.location,
+          navigationMode,
+          selectorMode: useSelectorMode
+        }
+      );
+    }
     const workflow = await this.workflowsCoreService.create({
       urls,
       name,
@@ -4654,7 +4702,8 @@ var ExtractionBuilderService = class {
       interval: this._options.interval,
       schedules: this._options.schedules,
       additionalData: this._options.additionalData,
-      bypassPreview: this._options.bypassPreview
+      bypassPreview: this._options.bypassPreview,
+      userPrompt: this._userPrompt
     });
     if (this._notificationOptions) {
       await this.notificationSetupService.setup({
@@ -5178,7 +5227,7 @@ var WSS_API_URI = process.env.KADOA_WSS_API_URI ?? "wss://realtime.kadoa.com";
 var REALTIME_API_URI = process.env.KADOA_REALTIME_API_URI ?? "https://realtime.kadoa.com";
 
 // src/version.ts
-var SDK_VERSION = "0.18.1";
+var SDK_VERSION = "0.19.1";
 var SDK_NAME = "kadoa-node-sdk";
 var SDK_LANGUAGE = "node";
 
@@ -5783,6 +5832,48 @@ var WorkflowsCoreService = class {
   async create(input) {
     validateAdditionalData(input.additionalData);
     const domainName = new URL(input.urls[0]).hostname;
+    if (input.navigationMode === "agentic-navigation") {
+      if (!input.userPrompt) {
+        throw new KadoaSdkException(
+          "userPrompt is required when navigationMode is 'agentic-navigation'",
+          {
+            code: "VALIDATION_ERROR",
+            details: { navigationMode: input.navigationMode }
+          }
+        );
+      }
+      const agenticRequest = {
+        urls: input.urls,
+        navigationMode: "agentic-navigation",
+        name: input.name ?? domainName,
+        description: input.description,
+        userPrompt: input.userPrompt,
+        schemaId: input.schemaId,
+        entity: input.entity,
+        fields: input.fields,
+        bypassPreview: input.bypassPreview ?? true,
+        tags: input.tags,
+        interval: input.interval,
+        monitoring: input.monitoring,
+        location: input.location,
+        autoStart: input.autoStart,
+        schedules: input.schedules,
+        additionalData: input.additionalData
+      };
+      const response2 = await this.workflowsApi.v4WorkflowsPost({
+        createWorkflowBody: agenticRequest
+      });
+      const workflowId2 = response2.data?.workflowId;
+      if (!workflowId2) {
+        throw new KadoaSdkException(ERROR_MESSAGES.NO_WORKFLOW_ID, {
+          code: "INTERNAL_ERROR",
+          details: {
+            response: response2.data
+          }
+        });
+      }
+      return { id: workflowId2 };
+    }
     const request = {
       urls: input.urls,
       name: input.name ?? domainName,